DAB & transparency

[NookieBandit]

mdk777, Jesse_V, thank you for the support.

Jesse_V, the link you provided was very helpful. I now see the priorities the DAB is currently working on. Since the generic topic of this thread is DAB and transparency and now knowing that Bruce is the representative for the default team, I'd like to suggest the following additional topic the DAB may consider (with Bruce's approval):

I did a search on the topic of tax deductibility for the cost of the electricity used to run a donor's F@H systems. Comments tend to center around it being difficult and/or the cost to Stanford/PG to implement and support the program.

On my own time and dime, I consulted with a leading tax accounting firm (CPAs and/or tax attorneys) on the prospect of getting an IRS ruling on the deductibility of the energy costs to support Stanford's F@H effort, assuming Stanford/PG is an official IRS 501(c)(3) registered entity. Quite surprisingly, the universal opinion is it would be approved, provided:

1. The systems are 100% dedicated with the exception of maintenance intervals
2. Determine the power draw from a watt meter dedicated solely to measuring the system and logging it
3. Affirm the average annual $/KWH for powering the system and notifying the charitable organization of the $/KWH the donor consumes
4. The donor logs all completed work units
5. The donor obtains documentation from Stanford/PG that provides evidence of all the work units submitted during the tax year (January 1 to December 31 for most filers).

The first four are all within the control of the donor. In fact, my UPS system will log the power used, and I can use HFM.NET for capturing the work units completed.

The 5th condition would require Stanford/PG to issue a simple email of a PDF document in January of each year using IRS form 990 (“Noncash Contributions”, Schedule M, Line 25) all filled out with the list of all work units processed and the total estimated electricity cost of processing them. All the donor would have to do is enter their social security number and use the form as part of their tax return for that tax year.

Doing a quick calculation of what the tax deductibility would mean for the top 1,000 contributors (using the extremeoverclocking.com stats) and applying the average cost/processed work I incur (which includes Linux, Windows & several GPUs), results in an annual cost of just under $1 million per year in electricity expense, which would result in an average of about a $200,000 tax deduction for the top 1,000 donors, or about $200 per year per donor. I don't think it would be unreasonable to assume (under US tax laws) that the contribution from US donors would be multiples of the tax deduction of the top 1,000. For the top donors, the tax deduction would have a substantial impact on the taxes they pay each year. I'd also suggest that PG, through the $/KWH contributed, would end up as being one of the largest recipients of donations to Stanford of any department in the University, which would be quite a PR coup!

In the spirit of helping donors offset some of the costs of running their systems by enabling Stanford/PG to issue IRS form 990's to donors at the end of the year, I respectfully ask Bruce to raise this with the DAB as a discussion topic to determine the feasibility.

If a moderator believes new thread should be started that provides a consolidated discussion for default donor's to suggest topics for Bruce to raise with other DAB members, please move this to a separate discussion.

[Punchy]

While that is a nice laundry list, I can already anticipate the response: "lack of resources". Maybe it should be cut down to something a little more manageable.

I'm also not convinced that it would really address the root cause of all the recent angst on this forum. Regardless of how much advance notice donors get about future optimization directions, the end result of those optimizations (points) is probably not known until a month or so before beta testing.

Donors that are that concerned with optimizing their purchases might do well to watch the results of other scientific projects rather than leaving their F@H blinders on. It's been pretty clear for years that GPU computing has enormous potential. As a recent example, in early October, the HCC project at WCG added GPU support, and it appears that a high-end AMD GPU can produce as many PPD as 10 dual Xeon E5 systems. I don't think we can expect Dr. Pande to keep us informed of the technology in areas other than F@H.

[Punchy]

The tax discussion should be a separate thread, but I see a few items for discussion already: deduction/depreciation of the equipment cost, consideration of the heating/cooling effects, and the difficulty of Stanford providing a power estimate.

[Leonardo]

Please, let's not derail this thread.

That's a compliment to those that have steered us in a good discussion.

[JimF]

While that is a nice laundry list, I can already anticipate the response: "lack of resources". Maybe it should be cut down to something a little more manageable.

My impression exactly. It would require a bureaucracy that Stanford undoubtedly has not bargained for. But if the DAB can get it to fly, more power to them. Just don't get too disappointed if it doesn't.

I'm also not convinced that it would really address the root cause of all the recent angst on this forum. Regardless of how much advance notice donors get about future optimization directions, the end result of those optimizations (points) is probably not known until a month or so before beta testing.

Ditto. And I think there are a number of research groups working on different projects; it is probably hard to know which ones will run how fast on a given piece of hardware.

Donors that are that concerned with optimizing their purchases might do well to watch the results of other scientific projects rather than leaving their F@H blinders on. It's been pretty clear for years that GPU computing has enormous potential. As a recent example, in early October, the HCC project at WCG added GPU support, and it appears that a high-end AMD GPU can produce as many PPD as 10 dual Xeon E5 systems. I don't think we can expect Dr. Pande to keep us informed of the technology in areas other than F@H.

I think that is the most important point. There are now a variety of projects for both CPUs and GPUs. WCG would love to have the CPU power that some people are discarding. And both WCG/HCC and POEM run well on AMD cards, which is something of a new phenomenon. So rather than trying to channel PG into a given course (or divining what it will be), I expect that it will be more practical to match the equipment you have to the projects that can best use it. There are a variety of worthwhile projects.

[orion]

Donors that are that concerned with optimizing their purchases might do well to watch the results of other scientific projects rather than leaving their F@H blinders on. It's been pretty clear for years that GPU computing has enormous potential. As a recent example, in early October, the HCC project at WCG added GPU support, and it appears that a high-end AMD GPU can produce as many PPD as 10 dual Xeon E5 systems. I don't think we can expect Dr. Pande to keep us informed of the technology in areas other than F@H.

I think that is the most important point. There are now a variety of projects for both CPUs and GPUs. WCG would love to have the CPU power that some people are discarding. And both WCG/HCC and POEM run well on AMD cards, which is something of a new phenomenon. So rather than trying to channel PG into a given course (or divining what it will be), I expect that it will be more practical to match the equipment you have to the projects that can best use it. There are a variety of worthwhile projects.

Interesting, are you saying if you're unhappy with [email protected] and go to another non F@H project?

The people here have bought hardware to do F@H not MW@H, E@H or any other non F@H project.

Let’s just drive more donors away!

[mdk777]

I expect that it will be more practical to match the equipment you have to the projects that can best use it. There are a variety of worthwhile projects.

Yeah, the objective was to help FOLDING not loose donors.

I guess just admitting defeat, closing down and telling donors to go elsewhere is another novel approach.

[JimF]

Interesting, are you saying if you're unhappy with [email protected] and go to BOINC?

I was doing BOINC (SETI and WCG) before Folding. When I get unhappy with Folding, I will notify you.

[road-runner]

Thats what I swap to WCG when I have had enough here but always find my way back some how...

[Duboisi]

Donors that are that concerned with optimizing their purchases might do well to watch the results of other scientific projects rather than leaving their F@H blinders on. It's been pretty clear for years that GPU computing has enormous potential. As a recent example, in early October, the HCC project at WCG added GPU support, and it appears that a high-end AMD GPU can produce as many PPD as 10 dual Xeon E5 systems. I don't think we can expect Dr. Pande to keep us informed of the technology in areas other than F@H.

I think that is the most important point. There are now a variety of projects for both CPUs and GPUs. WCG would love to have the CPU power that some people are discarding. And both WCG/HCC and POEM run well on AMD cards, which is something of a new phenomenon. So rather than trying to channel PG into a given course (or divining what it will be), I expect that it will be more practical to match the equipment you have to the projects that can best use it. There are a variety of worthwhile projects.

Interesting, are you saying if you're unhappy with [email protected] and go to BOINC?

The people here have bought hardware to do F@H not MW@H, E@H or any other BOINC clients.

Let’s just drive more donors away!

I would say is a reminder to PG that there are many options of contributing computing resources to Cancer Research and Protein Folding. Poor communication is enough to drive FAH donors away.

[Ben_Lamb]

Nicely said NookieBandit, especially for a 1st post and 55m points is still a hefty contribution to the project no matter what current rigs can do.

I'm olny going to add one thing to this, to prove that PG can sometimes get things nearly right. Last year they gave us 2 months notice of the changes to BA units, now many people, myself included, were not happy with the changes but at least we were given fair warning with plenty of time and testing to prove what machines did and didn't work. They also said that such units would be subject to yearly review - now that year is up and i haven't heard so PG - consider this a gentle reminder - not that the BA wu need any changes, 8101 is still tough enough to exclude all but the most powerful rigs

Whilst not perfect this was an example of how things should be done and not the current "here's a beta WU, allegedly doing the same work as the smp version, with a pile of points that bear no resemblance to the SMP WU that it mirrors, now fold it"

Shhhh dont remind them for all our sakes - you cant possibly want bigadv deadlines tightened again and if so you must be folding on alien technology !

I just wish they would get a bloody move on and sort the points mess out EQUAL POINTS FOR EQUAL WORK and we might start getting somewhere. We have seen 8057s now so dont expect us to be happy with 5k and 2k work units that take half a day to calculate. I have better things to spend my hard earned cash on, than pay for the lecky to crunch these rotten eggs. Pande group owe it to the donors to straighten things out as I for one wont fold anymore dud units that arent fairly rewarded with suitable points as the game just isnt fun that way. Thats what the points system was brought in for to make it fun for donors and its not much fun at the moment.

[Nathan_P]

No alien technology here, just a pile of parts awaiting a damn mobo. Actually i don't think they can tighten the deadlines any more, after last years announcement there have not been enough advances in chip speed to open up the floodgates again and have nearly everyone able to fold them, Intel have the E5 xeon that still requires considerable investment to be able to do BA units, only the top end hex and most of the octo cores can complete on time, westmere xeons are in the same place that they were last year and dual quads just don't cut it. AMD wise, interlagos was lame and only slightly faster on 8101, piledriver is only just out so i haven't seen any stats yet for the opteron chips and in both camps use of the core hack is now virtually nil - which is what PG and responsible teams wanted.

[JimF]

Just to set the record straight, I consider Dr. Pande to be a genius, and don't know of any other project that has the grasp of technology and science that this one has. But things change, as they have to if the project is going take advantage of new hardware and software techniques. If my hardware becomes more suitable for another project, I can move things around. The points will guide me in future upgrades, so they will all get my support to the extent possible.

[Rattledagger]

Interesting, are you saying if you're unhappy with [email protected] and go to BOINC?

The people here have bought hardware to do F@H not MW@H, E@H or any other BOINC clients.

While a few does buy hardware solely to crunch on, would still guess majority of FAH-donors also uses their computers for other things there running FAH is only one of multiple factors steering their buying-choise. The effect of FAH can example be to buy an i7 instead of an i3 or i5.

Let’s just drive more donors away!

Well, Amd GPU-usage dropping by roughly 75% the last month does indicate some problems...

[Punchy]

My statement about watching other projects had nothing to do with suggesting that people leave F@H - it was a suggestion about other ways to project the future potential of various technologies, before they get implemented by PG (which is not often an early adopter).

I think a great part of the frustration at the moment is the lack of communication about things that are actually in the present and past tense. There has been plenty of time now since the beta of GPU QRB started to evaluate both the stability and the production, yet not a single word from PG. This is why so many people's plans are now on hold.